Fixed updating of covariance matrix when both variables

[pspp-builds.git] / src / data / category.c

diff --git a/src/data/category.c b/src/data/category.c
index 835c0907bcfc48cb40a5faebee9246be973352c8..1620bc7f7f9b3c27bbde71737eeb860d5c4f7cf0 100644 (file)
--- a/src/data/category.c
+++ b/src/data/category.c
@@ -34,13 +34,14 @@
 #include <stdlib.h>
 #include <string.h>
 
-#include <libpspp/alloc.h>
 #include <libpspp/message.h>
 #include "category.h"
 #include "value.h"
 #include "variable.h"
 
-#define CAT_VALUE_NOT_FOUND -2
+#include "xalloc.h"
+
+#define CAT_VALUE_NOT_FOUND -1
 
 #define N_INITIAL_CATEGORIES 1
 
@@ -57,6 +58,11 @@ struct cat_vals
                                           track of the number of
                                           values stored.
                                         */
+  size_t *value_counts; /* Element i stores the number of cases for which
+                          the categorical variable has that corresponding 
+                          value. This is necessary for computing covariance
+                          matrices.
+                        */
 };
 
 void
@@ -69,6 +75,7 @@ cat_stored_values_create (const struct variable *v)
       obs_vals->n_categories = 0;
       obs_vals->n_allocated_categories = N_INITIAL_CATEGORIES;
       obs_vals->vals = xnmalloc (N_INITIAL_CATEGORIES, sizeof *obs_vals->vals);
+      obs_vals->value_counts = xnmalloc (N_INITIAL_CATEGORIES, sizeof *obs_vals->value_counts);
       var_set_obs_vals (v, obs_vals);
     }
 }
@@ -79,7 +86,10 @@ cat_stored_values_destroy (struct cat_vals *obs_vals)
   if (obs_vals != NULL)
     {
       if (obs_vals->n_allocated_categories > 0)
-        free (obs_vals->vals);
+       {
+         free (obs_vals->vals);
+         free (obs_vals->value_counts);
+       }
       free (obs_vals);
     }
 }
@@ -107,15 +117,17 @@ cat_value_find (const struct variable *v, const union value *val)
 }
 
 /*
-   Add the new value unless it is already present.
+   Add the new value unless it is already present. Increment the count.
  */
 void
 cat_value_update (const struct variable *v, const union value *val)
 {
   if (var_is_alpha (v))
     {
+      size_t i;
       struct cat_vals *cv = var_get_obs_vals (v);
-      if (cat_value_find (v, val) == CAT_VALUE_NOT_FOUND)
+      i = cat_value_find (v, val);
+      if (i == CAT_VALUE_NOT_FOUND)
        {
          if (cv->n_categories >= cv->n_allocated_categories)
            {
@@ -123,12 +135,36 @@ cat_value_update (const struct variable *v, const union value *val)
              cv->vals = xnrealloc (cv->vals,
                                    cv->n_allocated_categories,
                                    sizeof *cv->vals);
+             cv->value_counts = xnrealloc (cv->value_counts, cv->n_allocated_categories,
+                                           sizeof *cv->value_counts);
            }
          cv->vals[cv->n_categories] = *val;
+         cv->value_counts[cv->n_categories] = 1;
          cv->n_categories++;
        }
+      else
+       {
+         cv->value_counts[i]++;
+       }
     }
 }
+/*
+  Return the count for the sth category.
+ */
+size_t
+cat_get_category_count (const size_t s, const struct variable *v)
+{
+  struct cat_vals *tmp;
+  size_t n_categories;
+
+  tmp = var_get_obs_vals (v);
+  n_categories = cat_get_n_categories (v);
+  if (s < n_categories)
+    {
+      return tmp->value_counts[s];
+    }
+  return CAT_VALUE_NOT_FOUND;
+}
 
 const union value *
 cat_subscript_to_value (const size_t s, const struct variable *v)